Novel strategy for low-temperature, high-rate
growth of dense, hard, and stress-free
refractory ceramic thin films
Grzegorz Greczynski, Jun Lu, Stephan Bolz, Werner Koelker, Christoph Schiffers, Oliver
Lemmer, Ivan Petrov, Joseph E Greene and Lars Hultman
Linköping University Post Print
N.B.: When citing this work, cite the original article.
Original Publication:
Grzegorz Greczynski, Jun Lu, Stephan Bolz, Werner Koelker, Christoph Schiffers, Oliver
Lemmer, Ivan Petrov, Joseph E Greene and Lars Hultman, Novel strategy for low-temperature,
high-rate growth of dense, hard, and stress-free refractory ceramic thin films, 2014, Journal of
Vacuum Science & Technology. A. Vacuum, Surfaces, and Films, (32), 4, 041515.
http://dx.doi.org/10.1116/1.4884575
Copyright: American Vacuum Society
http://www.avs.org/
Postprint available at: Linköping University Electronic Press
refractory ceramic thin films
Grzegorz Greczynski, Jun Lu, Stephan Bolz, Werner Kölker, Christoph Schiffers, Oliver Lemmer, Ivan Petrov,
Joseph E. Greene, and Lars Hultman
Citation: Journal of Vacuum Science & Technology A 32, 041515 (2014); doi: 10.1116/1.4884575
View online: http://dx.doi.org/10.1116/1.4884575
View Table of Contents: http://scitation.aip.org/content/avs/journal/jvsta/32/4?ver=pdfcov
Published by the AVS: Science & Technology of Materials, Interfaces, and Processing
Articles you may be interested in
Mechanical behavior of mesoporous titania thin films
Appl. Phys. Lett. 104, 241902 (2014); 10.1063/1.4883260
Toughness enhancement in hard ceramic thin films by alloy design
APL Mat. 1, 042104 (2013); 10.1063/1.4822440
Nanomechanical properties of sputter-deposited HfO2 and HfxSi1-xO2 thin films
J. Appl. Phys. 110, 043527 (2011); 10.1063/1.3627155
Modulation periodicity dependent structure, stress, and hardness in NbN/W2N nanostructured multilayer films
J. Appl. Phys. 109, 123525 (2011); 10.1063/1.3598083
X-ray diffraction study of stress relaxation in cubic boron nitride films grown with simultaneous medium-energy ion bombardment
and stress-free refractory ceramic thin films
Grzegorz Greczynskia)and Jun Lu
Department of Physics (IFM), Link€oping University, SE-581 83 Link€oping, Sweden
Stephan Bolz, Werner K€olker, Christoph Schiffers, and Oliver Lemmer
CemeCon AG, Adenauerstr. 20 A4, D-52146 W}urselen, Germany
Ivan Petrov and Joseph E. Greene
Department of Physics (IFM), Link€oping University, SE-581 83 Link€oping, Sweden and Department of Materials Science, Physics, and the Frederick Seitz Materials Research Laboratory, University of Illinois, Urbana, Illinois 61801
Lars Hultman
Department of Physics (IFM), Link€oping University, SE-581 83 Link€oping, Sweden
(Received 23 January 2014; accepted 9 June 2014; published 24 June 2014)
Growth of fully dense refractory thin films by means of physical vapor deposition (PVD) requires elevated temperatures Ts to ensure sufficient adatom mobilities. Films grown with no external
heating are underdense, as demonstrated by the open voids visible in cross-sectional transmission electron microscopy images and by x-ray reflectivity results; thus, the layers exhibit low nanoindentation hardness and elastic modulus values. Ion bombardment of the growing film surface is often used to enhance densification; however, the required ion energies typically extract a steep price in the form of residual rare-gas-ion-induced compressive stress. Here, the authors propose a PVD strategy for the growth of dense, hard, and stress-free refractory thin films at low temperatures; that is, with no external heating. The authors use TiN as a model ceramic materials system and employ hybrid high-power pulsed and dc magnetron co-sputtering (HIPIMS and DCMS) in Ar/N2 mixtures to grow dilute Ti1xTaxN alloys on Si(001) substrates. The Ta target
driven by HIPIMS serves as a pulsed source of energetic Taþ/Ta2þ metal–ions, characterized by in-situ mass and energy spectroscopy, while the Ti target operates in DCMS mode (Ta-HIPIMS/Ti-DCMS) providing a continuous flux of metal atoms to sustain a high deposition rate. Substrate biasVsis applied in synchronous with the Ta-ion portion of each HIPIMS pulse in
order to provide film densification by heavy-ion irradiation (mTa¼ 180.95 amu versus
mTi¼ 47.88 amu) while minimizing Arþbombardment and subsequent trapping in interstitial sites.
Since Ta is a film constituent, primarily residing on cation sublattice sites, film stress remains low. Dense Ti0.92Ta0.08N alloy films, 1.8 lm thick, grown withTs 120C (due to plasma heating) and
synchronized bias,Vs¼ 160 V, exhibit nanoindentation hardness H ¼ 25.9 GPa and elastic modulus
E¼ 497 GPa compared to 13.8 and 318 GPa for underdense Ti-HIPIMS/Ti-DCMS TiN reference layers (Ts< 120C) grown with the sameVs, and 7.8 and 248 GPa for DCMS TiN films grown with
no applied bias (Ts< 120C). Ti0.92Ta0.08N residual stress is low, r¼ 0.7 GPa, and essentially
equal to that of Ti-HIPIMS/Ti-DCMS TiN films grown with the same substrate bias. VC 2014
American Vacuum Society. [http://dx.doi.org/10.1116/1.4884575] I. INTRODUCTION
The quest for low-temperature processes for the growth of dense, hard, low-stress refractory thin films has been a recurring theme in materials science for many decades.1–10 The use of inert-gas ion irradiation,1,2,7to provide dynamic ion mixing and ion-bombardment-enhanced surface adatom mobilities3,7,11 during physical vapor deposition, is widely employed for improving film density. However, progress achieved in low temperature densification often comes at the price of incorporating large compressive stresses due to trap-ping of inert gas ions, and recoil implantation of surface atoms, into film interstitial sites.3,11–18
Here, we demonstrate a new strategy for synthesizing high-hardness, low-stress ceramic thin films at low growth tempera-tures (Ts 120C) based upon reactive hybrid high-power
pulsed and dc magnetron co-sputtering (HIPIMS/DCMS) with synchronized substrate bias19–21with synchronized substrate bias in which a small concentration of a high-mass metal ele-ment, with a low first-ionization potential IP1, is provided
during HIPIMS pulses. As an example of this approach, we deposit cubic NaCl-structure TiN by high-rate DCMS from a Ti (47.88 amu) target and add TaN to form a dilute solid so-lution (TiN and TaN exhibit complete solid solubility)22 using HIPIMS/DCMS co-sputtering from a Ta (180.95 amu) target in mixed Ar/N2discharges. A negative substrate bias
Vsis applied in synchronous with the metal–ion portion (2%
duty cycle) of each HIPIMS pulse. The growing film is at floating potential, Vf¼ 10 V, between HIPIMS pulses. The
a)
Electronic mail: grzgr@ifm.liu.se
relatively low first and second ionization potentials of Ta (IP1¼ 7.91 and IP2¼ 15.60 eV)23,24compared with those of
Ar (IP1¼ 15.76 eV and IP2¼ 27.63 eV),
25
the principal dis-charge gas, provides significant Taþ ion fluxes to the growth surface, as determined byin-situ mass spectroscopy, resulting in film densification by generating large numbers of low-energy N and Ti primary recoils with significant cas-cade overlap leading to efficient ion-irradiation-induced mixing.
Film microstructure evolves from underdense with both intra- and intercolumnar porosity, as determined by cross-sectional transmission electron microscopy (XTEM), scan-ning electron microscopy (XSEM), and x-ray reflectivity (XRR) results for TiN and Ti0.92Ta0.08N reference layers
de-posited at low Ts with no Ta ion irradiation, to essentially
fully dense for Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N
poly-crystalline layers grown at low Ts with synchronized Vs
¼ 160 V Ta ion irradiation. In addition, the hardness and elastic modulus of Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N
alloys increase with increasing substrate bias from 14.4 and 346 GPa with Vs¼ 20 V to 25.9 and 497 GPa with
Vs¼ 160 V. Since Ta ions are film constituents, layer stress
remains low (r 0.7 GPa). Moreover, due to the low substrate-bias duty cycle, heating generated by pulsed Tanþ ion bombardment is negligible. The deposition rate is main-tained high by continuous DCMS sputtering of the Ti target. Further increases inVs> 160 V have no significant effect on
H and E, but film compressive stress increases. II. EXPERIMENT
Ti1xTaxN films are grown in a CC800/9 CemeCon AG
magnetron sputtering system26equipped with cast rectangu-lar 8.8 50 cm2Ti and Ta targets. Si(001) substrates, 2 1 cm2, are mounted symmetrically with respect to the targets, which are tilted toward the substrate, resulting in a 21angle between the substrate normal and the normal to each target. Target-to-substrate distance is 18 cm, and the system base pressure is 3.8 107Torr (0.05 mPa). The substrates are cleaned sequentially in acetone and isopropyl alcohol and mounted with clips such that their long sides are parallel to the long sides of the targets. The growth chamber is degassed before deposition by applying 0.5 kW to each of two resistive heaters for 1 h, resulting in a chamber temperature of 110C at the substrate position; the power is then switched off 1 h prior to the start of film growth such that the substrate temper-atureTsdrops to 65C.Tsis monitored during the deposition
process with a calibrated27thermocouple attached to a sacrifi-cial dummy substrate. The Ar flow rate is set at 350 cm3/min, while the N2flow is controlled by an automatic pressure
regu-lator via a feedback loop to maintain the total pressure Ptot
constant during deposition at 3 mTorr (0.4 Pa).
A hybrid target power scheme is employed in which the Ta magnetron is operated in HIPIMS mode to supply a pulsed Ta-ion flux, while the Ti target is operated as a con-ventional dc magnetron (Ta-HIPIMS/Ti-DCMS).19The Taþ and Ta2þ metal–ion fluxes incident at the film growth sur-face, and hence, the TaN concentration in Ti1xTaxN films,
is controlled by varying the average power to the HIPIMS target from 0.5 to 1.5 kW, at a fixed pulsing frequency of 100 Hz (2% duty cycle). This corresponds to varying the pulse energyEpfrom 5 to 15 J. The peak target current
den-sityJTduring the HIPIMS pulse is 0.21 A/cm2withEp¼ 5 J,
and increases to 0.53 and 0.93 A/cm2 withEpvalues of 10
and 15 J.28 Resulting TaN concentrations range from x¼ 0.08 (Ep¼ 5 J) to 0.16 (Ep¼ 15 J) as determined by
energy-dispersive x-ray spectroscopy (EDS). The power to the dc magnetron is maintained constant at 6 kW correspond-ing to a TiN deposition rate of 277 A˚ /min. Deposition rates for Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08 N films range from
340 A˚ /min with Vs¼ 20 V to 295 A˚ /min with Vs¼ 280 V.
A pulsed substrate biasVsis synchronized with the 200 ls
metal–ion-rich portion of each HIPIMS pulse beginning at toffset¼ 80 ls after pulse initiation (t ¼ 0 ls), as determined
by time-resolved ion mass spectroscopy analyses at the sub-strate position. At all other times, the subsub-strate is at a nega-tive floating potential Vs¼ Vf¼ 10 V. For films grown with
Ep¼ 5 J, Vsis varied from 20 to 280 V in order to probe the
effects of incident Ta ion energy on film nanostructure. The total film deposition time is 60 min, and the resulting film thicknesses range from 2.04 lm atVs¼ 20 V to 1.77 lm
with Vs¼ 280 V. In addition, Ti1xTaxN layers grown with
Ep¼ 10 J and Vs varied from 20 to 80 V, as well as with
Ep¼ 15 J and Vsvaried from 20 to 60 V, are synthesized to
investigate the effect of higher Taþ/Ta2þ metal–ion fluxes on film properties.
TiN/Ti0.92Ta0.08N/TiN trilayer stacks with total
thick-nesses of3 lm are used for comparative XTEM nanostruc-tural analyses. The DCMS TiN under- and overlayers are grown with Vs¼ 10 V, and a metal–ion-synchronized
sub-strate bias of Vs¼ 160 V is applied during growth of the
Ti0.92Ta0.08N intermediate layer.
Pure DCMS TiN and Ti0.92Ta0.08N reference films are
de-posited by conventional dc magnetron sputtering at an aver-age power of 6 kW and a floating potential Vs¼ 10 V with
no intentional heating (Ts< 120C). In addition, a DCMS
TiN reference film is grown with continuous applied bias Vs¼ 60 V at Ts¼ 500C. Ti-HIPIMS/Ti-DCMS TiN
refer-ence layers are also grown, following the same procedure as for the Ta-HIPIMS/Ti-DCMS layers (pulsing frequency of 100 Hz, 2% duty cycle, Ep¼ 5 J, and JT¼ 0.30 A/cm
2
), except that the growing films are subjected to Tiþ/Ti2þ, rather than Taþ/Ta2þ, irradiation initiated at toffset¼ 40 ls
during the synchronously biased metal–ion portion of each HIPIMS pulse.
Time-dependent in-situ mass and energy spectroscopy analyses of the ion fluxes incident at the substrate for Ta and Ti targets operated in HIPIMS mode are carried out using a Hiden Analytical EQP1000 instrument to determine the composition, charge state, and energy of ion fluxes incident at the growing film as a function ofEp. In these experiments,
the orifice of the mass spectrometer is placed parallel to the target surface at the substrate position. Ion-energy-distribu-tion funcIon-energy-distribu-tions (IEDFs) are recorded in HIPIMS mode for Arþ, Ar2þ, Nþ, N2þ, Taþ, Ta2þ, Tiþ, and Ti2þwhile
pressure and gas flow rates as used for the film growth experiments. More details can be found in Ref.29.
Ti1xTaxN film compositions x are determined by
EDS measurements carried out on fracture cross-sections. N/(Tiþ Ta) fractions are obtained by Rutherford backscat-tering spectroscopy (RBS) using a 2.0 MeV 4Heþ probe beam incident at 10 with respect to the surface normal and detected at a 172scattering angle. Experimental data are an-alyzed based uponSIMNRA6.06 software.30Film thicknesses
are determined from XSEM analyses in a LEO 1550 instrument.
h-2h x-ray diffraction (XRD) scans in 2h steps of 0.1,
XRR, and sin2w measurements for residual stress determina-tions,31 are carried out using a Philips X’Pert MRD system operated with point-focus Cu Ka radiation. Relaxed TiN and Ti0.92Ta0.08N lattice parametersaoare determined from h-2h
scans acquired at the strain-free tilt angle w defined as w¼ arcsin½ 2=ð1 þ Þð Þ1=2,19 in which is the Poisson ratio.
The sin2w method employs Hooke’s law of linear elastic-ity expressed as31
e wð Þ ¼1þ E r sin
2w2
E r; (1)
for which is the Poisson ratio and e is the film strain obtained from
e wð Þ ¼dw d0 d0
: (2)
Lattice spacingsdw in Eq.(2)are determined for each
sam-ple from the positions of the cubic 111 Bragg reflections at ten different values of the tilt angle w ranging from 0 to 71.57, with w steps that produce equally spaced data points on the sin2w axis. The relaxed interplanar spacing do is
acquired at the strain-free tilt angle w*;31 elastic moduli E inserted in Eq. (1) are obtained from nanoindentation measurements (Fig.7); and Poisson ratios, ¼ 0.25 for TiN and 0.23 for Ti0.92Ta0.08N, are from Refs. 32 and 33,
respectively.
Plan-view transmission electron microscopy (TEM) and XTEM specimens are prepared by mechanical polishing, fol-lowed by several hours of Arþion milling at 5 kV with an 8 incidence angle and sample rotation. During the final stages (10–20 min) of sample thinning, the ion energy and inci-dence angle are reduced to 2.5 kV and 5. Film nanostruc-tures are analyzed in an FEI Tecnai G2 TF 20 UT transmission electron microscope operated at 200 kV.
X-ray photoelectron spectroscopy (XPS) compositional depth profiles of as-deposited films are acquired in a Axis Ultra DLD instrument from Kratos Analytical using monoochromatic Al Ka radiation (h¼ 1486.6 eV). The results serve as a qualitative measure of film density and are consistent with XTEM analyses, since highly underdense layers exhibit significant oxygen concentrations throughout the film following air exposure. Arþsputter etching at 4 keV and 12.7 mA/cm2, an incident angle of 70 with respect to the sample normal, and
with the beam rastered over a 3 3 mm2area, is performed in one-minute steps, corresponding to the removal of 160 A˚ /step as determined by XSEM fracture cross-section calibrations. The area analyzed by XPS is 0.3 0.7 mm2and centered in the middle of the ion-etched crater.
A sharp Berkovich diamond tip is used to determine nanoindentation hardnessH and elastic moduli E of all films, both reference samples and Ta-HIPIMS/Ti-DCMS layers, following the procedure of Oliver and Pharr.34A minimum of 20 indents, with a maximum load of 15 mN, are made in each sample. Indentation depths range from 1500 to 2000 A˚ , but are never allowed to exceed 10% of the film thickness in order to minimize substrate effects.
Transport of ions in matter (TRIM),35 a Monte Carlo
pro-gram included in the stopping power and range of ions in matter (SRIM) software package,36is used to estimate
primary-ion and recoil projected ranges due to Tanþand Tinþmetal– ion irradiation during Ti1xTaxN and TiN film growth.
III. RESULTS
A. Mass and energy spectroscopy analyses
In situ mass spectrometry and energy spectroscopy meas-urements reveal the presence of large fluxes of energetic Taþ and Ta2þmetal–ions incident at the substrate position during Ta-HIPIMS pulses. Figures 1(a) and 1(b) show Taþ and Ta2þ IEDFs acquired during 200 ls metal–ion HIPIMS pulses (fromtoffset¼ 80 ls to t ¼ 280 ls) with energy Ep
var-ied from 3 to 18 J in steps of 3 J. For singly ionized Taþ, the primary effect of increasingEpis an increase in the average
ion energy, from 4.9 eV with Ep¼ 3 J to 9.9 eV with Ep
¼ 18 J, due to increasing intensities in the high-energy tails. The maximum Taþion intensity increases from 1.2 106to 3.2 106cps. The average energy of Ta2þions ranges from 7.6 eV withEp¼ 3 J to 8.5 and 9.8 eV with Ep¼ 12 and 18 J
as the maximum flux intensity increases from 3.5 105 to 4.1 106and 3.1 106cps. The doubly ionized Ta2þ frac-tion Ta2þ/(Taþþ Ta2þ) in the metal–ion flux incident at the film growth surface during HIPIMS pulses increases from 0.33 with Ep¼ 3 J to 0.45 at 18 J. With Ep¼ 5 J, the
pulse energy used for growth of Ti0.92Ta0.08N films,
Ta2þ/(Taþþ Ta2þ)¼ 0.36.
Time-resolved measurements during HIPIMS pulses show that with a pulse energyEp¼ 5 J, the first Taþions reach the
substrate plane at t¼ 60 ls with a mean energy of 20 eV, corresponding to a flight time of 28 ls. Thus, sputtering of the Ta target begins at t 30 ls after pulse initiation. The metal–ion flux at the substrate increases with time into the pulse and at t¼ 80 ls exceeds that of the inert-gas ions. The maximum Taþflux JTaþ is obtained neart¼ 130 ls. At
longer times, JTaþ decreases slowly until at t > 280 ls,
inert-gas ions again dominate the ion flux to the substrate. Doubly charged Ta2þions are predominantly produced dur-ing the most energetic phase of the discharge. JTa2þ at the
substrate reaches a maximum neart¼ 120 ls.
Corresponding Tiþ and Ti2þ metal-ion fluxes incident at the substrate position during Ti-HIPIMS/Ti-DCMS TiN film growth are shown for reference in Figs. 1(c) and1(d).
Tinþand TanþIEDF shapes differ significantly. This is par-ticularly obvious at Ep 12 J, for which high-intensity
peaks, with relatively narrow energy distributions, appear at low energies superimposed on the broad IEDFs. The low-energy peaks arise due to Tiþand Ti2þions being more efficiently thermalized than their high-mass Taþ/Ta2þ coun-terparts during transport through the plasma. That is, the mass match between Ti (47.87 amu) and Ar (39.95 amu) is much better than for Ta (180.95) and Ar. The Ti2þfraction Ti2þ/(Tiþþ Ti2þ) in the metal-ion flux incident at the grow-ing film surface ranges from 0.08 withEp¼ 3 J to 0.30 with
Ep¼ 18 J. Ti-HIPIMS/Ti-DCMS TiN reference layers are
grown withEp¼ 5 J, for which Ti2þ/(Tiþþ Ti2þ)¼ 0.12.
Previous mass spectroscopy results for Ti-DCMS, under the same deposition conditions as employed here, reveal that the ion flux incident at the growing film is dominated by low-energy gas species (primarily Arþand N2þ).19Average
ion energies are 2.3 eV for Arþ(84% of the ion flux) and 2.8 eV for N2þ (4.9%). Tiþ ions have higher energies,
4.8 eV, but account for only 10% of the ion flux incident at the substrate plane, while the Ar2þ, Ti2þ, and Nþion fluxes are slightly higher than the detection limit and constitute 0.8, 0.2, and 0.1% of the total ion flux, respectively.
B. Ti12xTaxN film composition
The concentrationx of TaN in Ti1xTaxN films increases
from x¼ 0.08 with Ep¼ 5 J, to 0.13 and 0.16 with Ep¼ 10
and 15 J, respectively. Ti1xTaxN alloys grown withEp¼ 5 J
and substrate biasesVs¼ 20 – 280 V applied in synchronous
with the metal–ion rich portions of Ta-HIPIMS pulses are found to be of constant composition with x¼ 0.08. RBS analyses indicate that N/(Taþ Ti) ¼ 1.00 6 0.03 for all Ti1xTaxN alloys.
C. Nanostructure
As shown in Secs.III DandIII E, the optimum combina-tion of high hardness and elastic modulus in fully dense films with essentially zero residual stress is obtained for Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N layers grown with
Vs¼ 160 V. Thus, we primarily concentrate on these alloys
for comparison to TiN reference layers grown by DCMS and Ti-HIPIMS/Ti-DCMS.
Figure2shows three sets of w-2h scans as a function of the tilt angle w, defined as the angle between the surface nor-mal and the diffraction plane containing the incoming and diffracted x-ray beams, for (a) DCMS TiN films deposited with Vs¼ 10 V, (b) Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N
films grown with metal–ion-synchronized substrate bias Vs¼ 160 V, and (c) Ti-HIPIMS/Ti-DCMS TiN layers grown
as in (b) withVs¼ 160 V. There is no applied heating in the
three sets of growth experiments and the maximum tempera-ture during deposition, due to plasma heating, is 120C
(see the Appendix). The only peaks observed in the w-2h scans are cubic NaCl-structure 111 and 002 reflections. Thus, we focus in Fig. 2on the 2h region between 33 and 47, which includes both peaks. The tilt angle w is varied from 0 to 71.6such that sin2w ranges from 0 to 0.9 in steps of 0.1.
For DCMS TiN films, the 002 peak position obtained at the strain-free tilt angle, w*¼ 39.2, is 42.51corresponding
to a relaxed lattice parameterao¼ 4.25 6 0.01 A˚ , essentially
equal to TiN powder diffraction data.37 The 111 and 002 peak positions do not change significantly as a function of the tilt angle w, indicative of low residual stress. Normalizing peak intensities to those obtained from powder diffraction,37reveals that DCMS TiN films have very strong 111 preferred orientation.
FIG. 1. (Color online) IEDFs for: (a) Taþ, (b) Ta2þ, (c) Tiþ, and (d) Ti2þmetal–ions, incident at the substrate position, from Ta and Ti targets powered by HIPIMS as a function of pulse energyEp.
The 111 and 002 diffraction peaks from Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N alloys are broader [see Fig.2(b)],
ex-hibit reduced intensity, and are shifted to lower 2h values compared to DCMS TiN layer results in Fig. 2(a). The strain-free (w¼ 39.2) Ti
0.92Ta0.08N 002 XRD peak position
is 42.24, compared to 42.51for DCMS TiN, resulting in a larger relaxed lattice parameter,ao¼ 4.28 6 0.01 A˚ , which is
independent of the synchronous substrate biasVs. As was the
case for DCMS TiN, Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N
111 and 002 peak positions do not shift significantly as a function of w, indicative of low film stresses; the alloys ex-hibit 111 preferred orientation.
The Ti-HIPIMS/Ti-DCMS TiN film strain-free 002 XRD peak position in Fig.2(c)is 42.63 yielding a relaxed lattice
parameter ao¼ 4.24 6 0.01 A˚ , close to that of DCMS TiN
(4.25 6 0.01 A˚ ). These films also exhibit low residual stresses with 111 preferred orientation.
1. TiN/Ti0.92Ta0.08N/TiN trilayers
Figure 3 is a bright-field XTEM image of a trilayer (DCMS TiN)/(Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N)/(DCMS
TiN) sample grown on a Si(001) wafer without external heat-ing. The layer thicknesses are 8700, 9300, and 8250 A˚ ; the Ti0.92Ta0.08N alloy layer is deposited with Ep¼ 5 J; and a
metal–ion-synchronized substrate bias Vs¼ 160 V. DCMS
TiN layers are grown with the substrate electrically floating,
FIG. 2. (Color online) w-2h scans for: (a) DCMS TiN films grown with Vs¼ 10 V, (b) Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N films grown with
Vs¼ 160 V, and (c) Ti-HIPIMS/Ti-DCMS TiN layers with Vs¼ 160 V. For
both (b) and (c), the bias is synchronized with the metal–ion rich portions of HIPIMS pulses. In all three cases, there is no external substrate heating (Ts 120C) during film growth on Si(001) substrates. The vertical lines in
each panel designate the positions of TiN reflections from reference powder diffraction data (Ref.37).
FIG. 3. (a) Bright-field XTEM image of a trilayer (DCMS
TiN)/(Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N)/(DCMS TiN) film grown without external
heating. The Si(001) substrate is electrically floating, Vs¼ 10 V, during
TiN deposition; while for alloy growth, Vs¼ 160 V synchronized to the
metal–ion-rich portions of HIPIMS pulses. SAED patterns from each layer are shown as insets. Higher-resolution XTEM images are also presented highlighting (b) the interface between the lower DCMS TiN layer and the Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N alloy, (c) the interface between the
Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N alloy film and the upper DCMS TiN
film, and (d) the top surface of the trilayer sample.
Vs¼ Vf¼ 10 V. Image contrast between Ti0.92Ta0.08N and the
two TiN encapsulation layers is due to enhanced electron absorption by the heavier Ta atoms combined with the higher density (see below) of the Ta-HIPIMS/Ti-DCMS film.
The DCMS TiN underlayer in Fig. 3 has a columnar nanostructure and pronounced intercolumnar voids. The av-erage column diameter is 100 6 20 A˚ near the substrate and increases to 570 6 160 A˚ at the top of the layer. The columns also exhibit an intracolumnar network of voids [see Fig.3(b)], with typical size <10 A˚ . This is a signature of low adatom surface mobilities.3,38 XTEM and XSEM images reveal that the 111-oriented TiN columns have conical upper surfaces with sharp tips and an average peak-to-valley dis-tancew of 300 6 100 A˚ . From plan-view images, the cones are approximately symmetric. The layer roughness results from kinetic roughening due to the presence of step-edge Ehrlich barriers39,40 on locally epitaxial column surfaces. The roughening rate is exacerbated by atomic shadowing arising from the combination of an approximately cosine point emission pattern from the target41–44 and gas-phase scattering.45
The increase in film density between the lower TiN and middle Ti0.92Ta0.08N layer is clearly shown in the
higher-resolution XTEM image in Fig. 3(b) and consistent with XRR results in Sec. III C 2. Columnar growth is preserved; the average column diameter is 570 6 120 A˚ at the bottom of the alloy layer, close to the average column size in the upper region of the underlying DCMS TiN layer, suggesting local epitaxy. The intracolumnar voids disappear within the first 100 A˚ into the Ti0.92Ta0.08N layer due to pulsed Taþ/Ta2þ
metal–ion irradiation-induced atomic mixing. However, the intercolumnar voids persist for 500–1000 A˚ because of low adatom mean-free-paths compared to average feature sizes. Figure 3(c) reveals that the Ti0.92Ta0.08N surface is much
smoother than that of the TiN underlayer; the column tops have become rounded and shallower with an average peak-to-valley distance that has decreased by approximately a factor of two tow¼ 150 6 50 A˚ . The average column diame-ter at the top of the 9300-A˚ -thick Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N interlayer is 1980 6 200 A˚ , an increase by a
factor of 3.5.
Figure 3(c)also shows that intercolumnar porosity reap-pears, within less than 50 A˚ , during growth of the upper
DCMS TiN layer with Vs¼ 10 V. The rough surface and
associated intercolumnar porosity observed in the DCMS TiN underlayer [Figs. 3(a)and3(b)] also develop as shown in Fig.3(d), an XTEM image of the surface region of the tri-layer film. The average column diameter of the 8250 A˚ -thick upper DCMS TiN film is 80 6 20 A˚ near the interface with the Ti0.92Ta0.08N interlayer and increases to 680 6 260 A˚ in
the surface region. The average peak-to-valley roughness is 500 6 300 A˚ .
Corresponding selected area electron diffraction (SAED) patterns acquired from the middle portions of each layer of the TiN/Ti0.92Ta0.08N/TiN sample are shown as insets in Fig. 3. All layers exhibit 111 texture in agreement with the XRD results presented above.
2. Single layers
Figure4compares fracture XSEM images acquired near the top of a 1.95 lm-thick DCMS TiN reference film depos-ited with Vs¼ 10 V [Fig. 4(a)] and a 1.85 lm-thick
Ti0.92Ta0.08N Ta-HIPIMS/Ti-DCMS alloy layer grown with
a metal–ion-synchronized bias Vs¼ 160 V [Fig. 4(b)]. In
both cases, Ts 120C (see the Appendix). The TiN
refer-ence sample is characterized by conical column tops with an average peak-to-valley distancew¼ 700 A˚ ; the film fractures in an intercolumnar manner and the side walls of the pillars exhibit nanostructure, which corresponds to nanoporosity within the columns as observed by XTEM [Fig. 3(d)]. The surfaces of the Ti0.92Ta0.08N Ta-HIPIMS/Ti-DCMS films are
much flatter exhibiting mounded structures with heights 50–100 A˚ , shallow troughs at boundaries, and no discernable porosity. The cross-sectional image reveals mixed inter- and intracolumnar fracture.
Figure5shows XPS oxygen concentration depth profiles CO(h) acquired over the first 5000 A˚ from the surface of
air-exposed layers: (a) a Ref. DCMS TiN film grown with Vs¼ Vf¼ 10 V, (b) a Ti-HIPIMS/Ti-DCMS TiN layer, and
(c) a Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N alloy layer. Both
films (b) and (c) were grown with a metal–ion synchronized bias of 160 V. CO for the DCMS TiN film, after removal
of the surface native oxide layer, saturates at a value of 8 at. % with hⲏ 1000 A˚. This is a sign of an underdense microstructure. The oxygen concentration profile for the Ti-HIPIMS/Ti-DCMS TiN layer deposited with
FIG. 4. Cross-sectional SEM images of the area near the top of (a) a DCMS TiN reference film grown on Si(001) withVs¼ 10 V and (b) a
Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N alloy layer grown on Si(001) withVs¼ 160 V synchronized to the metal-ion-rich portions of HIPIMS pulses. In both cases, there is no
synchronized Tiþ/Ti2þ metal–ion irradiation pulses is very similar to that of the DCMS TiN film, but with a lower bulk concentration of 5 at. %, indicative of lower porosity. In contrast, the bulk filmCOvalue for Ti0.92Ta0.08N alloy films
grown with synchronized Taþ/Ta2þmetal–ion bombardment pulses is less than the oxygen detection limit,1 at. %, indi-cating a dense nanostructure, consistent with cross-sectional electron microscopy images in Figs.3and4.
We quantify the density of representative single-layer films using XRR and a stoichiometric, single-crystal TiN/ MgO(001) reference sample.46DCMS TiN layers deposited with Ts< 120C and Vs¼ 10 V have a relative density of
65%. The density of Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N
layers grown with Vs¼ 20 V is 76% (relative to 6.14 g/cm 3
for a fully dense alloy film with 8 mol. % TaN and ao
¼ 4.28 A˚ ), while Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N layers
grown withVs¼ 160 V exhibit a density of 98%.
Optical photographs show that DCMS TiN reference films grown withVs¼ 10 V exhibit a dark brown appearance
typi-cal of films deposited at low temperature (Ts/Tm< 0.12 in this
case, whereTmis the melting temperature) and, hence, have
high surface roughness resulting in light trapping and low reflectivity. Ti0.92Ta0.08N Ta-HIPIMS/Ti-DCMS alloy layers
grown with metal–ion synchronized bias Vs¼ 160 V are
golden in color with a shiny reflective surface.
D. Nanoindentation hardness and elastic moduli
Nanoindentation hardness H and elastic moduli E of Ti0.92Ta0.08N films, deposited with no external heating
(Ts 120C), are plotted in Figs.6and7as a function of the
applied substrate bias voltageVssynchronized to the metal–
ion portions of the HIPIMS pulses. Results are also shown for four sets of reference layers: (1) DCMS TiN deposited with Ts< 120C and Vs¼ 10 V, (2) DCMS TiN grown at
Ts¼ 500C with continuously applied bias, Vs¼ 60 V, (3)
Ti-HIPIMS/Ti-DCMS TiN layers deposited at Ts 120C
with substrate bias 20 Vs 280 V synchronized to the
metal–ion rich portions of the HIPIMS pulses, and (4) DCMS Ti0.92Ta0.08N alloys grown at Ts< 120C with
Vs¼ 10 V.
H for DCMS TiN reference layers grown with no external heating andVs¼ 10 V is very low, 7.8 GPa, as expected due to
the underdense nanostructure (see Figs.3–5), and only mar-ginally higher, 9.8 GPa, for DCMS Ti0.92Ta0.08N alloy films.
The DCMS TiN layer grown at 500C has hardness H¼ 19.4 GPa, similar to values reported for epitaxial
FIG. 5. (Color online) XPS oxygen concentration depth profilesCO(h) from
the air-exposed surface through the first 5000 A˚ of (a) a reference DCMS TiN film grown with Vs¼ 10 V, (b) a Ti-HIPIMS/Ti-DCMS TiN layer
grown withVs¼ 160 V, and (c) a Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N alloy
layer grown withVs¼ 160 V. For the films in (b) and (c), the applied bias is
synchronized with the metal–ion-rich portions of HIPIMS pulses. In all three cases, there is no external substrate heating (Ts 120C) during film growth
on Si(001) substrates.
FIG. 6. (Color online) HardnessH of Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N/
Si(001) films as a function of the applied substrate bias voltageVs
synchron-ized to the metal-ion-rich portions of HIPIMS pulses. Results are also shown for four sets of reference layers grown on Si(001) substrates: (1) DCMS TiN deposited with Vs¼ 10 V, (2) DCMS TiN grown at Ts¼ 500C with
Vs¼ 60 V, (3) Ti-HIPIMS/Ti-DCMS TiN films deposited with synchronized
substrate bias 20 Vs 280 V, and (4) DCMS Ti0.92Ta0.08N alloys grown at
Ts< 120C withVs¼ 10 V. Except for TiN reference layer (2), there is no
external substrate heating andTs 120C.
FIG. 7. (Color online) Elastic moduli E of Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N/Si(001) films as a function of the applied substrate bias voltage
Vssynchronized to the metal-ion-rich portions of HIPIMS pulses. Results
are also shown for four sets of reference layers grown on Si(001) substrates: (1) DCMS TiN deposited with Vs¼ 10 V, (2) DCMS TiN grown at
Ts¼ 500C withVs¼ 60 V, (3) Ti-HIPIMS/Ti-DCMS TiN films deposited
with synchronized substrate bias 20 Vs 280 V, and (4) DCMS
Ti0.92Ta0.08N alloys grown atTs< 120C withVs¼ 10 V. Except for TiN
reference layer (2), there is no external substrate heating andTs 120C.
TiN(001).47,48 For Ti-HIPIMS/Ti-DCMS, TiN films grown withVs¼ 20 V, H is low and equal to 9.4 GPa. With increasing
Vs, Ti-HIPIMS/Ti-DCMS TiN films exhibit a linear increase
in hardness to 18.8 GPa atVs¼ 280 V. Ta-HIPIMS/Ti-DCMS
Ti0.92Ta0.08N layers grown withVs¼ 20 V have a hardness of
14.4 GPa, almost twice that of low-TsDCMS TiN. Increasing
Vs to 60, 80, and 120 V, results in further increases in H to
19.1, 21.8, and 24.1 GPa, respectively. At still higher synchronized biases, 160 Vs 280 V, H remains relatively
constant at 26 6 1.3 GPa.
The results presented in Fig.6indicate that even a small dose of heavy-metal–ion bombardment per HIPIMS pulse has a large effect on film properties. Synchronized Vs
¼ 160 V metal–ion bombardment with Taþ/Ta2þions during Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N deposition results in
layers with significantly higher hardness, 25.9 GPa, than that of Ti-HIPIMS/Ti-DCMS TiN films grown with Tiþ/Ti2þion irradiation pulses, 13.8 GPa, using the same HIPIMS condi-tions (5 J/p, 100 Hz, and 2% duty cycle). The hardness of Ti-HIPIMS/Ti-DCMS TiN layers grown with Vs¼ 20 V is
only 9.4 GPa, i.e., comparable to low-Ts TiN DCMS film,
both significantly lower than for Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N layers. With increasing synchronous substrate
bias, Ti-HIPIMS/Ti-DCMS TiN films exhibit a linear increase inH. However, over the entire substrate bias range, 20 Vs 280 V, Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N alloys
have hardness, which are 50–70% higher than Ti-HIPIMS/ Ti-DCMS layers.
The trends in elastic moduliE dependences on x and Vs(see
Fig.7) follow the hardness results described above. For DCMS TiN films, E increases from 248 GPa for films grown with Ts< 120C and Vs¼ 10 V to E ¼ 509 GPa with Ts¼ 500C
andVs¼ 60 V. The latter is close to previously reported values
for epitaxial TiN(001).47,48For DCMS, Ti0.92Ta0.08N films
de-posited with no external heating and floating bias,E¼ 286 GPa. Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N alloys grown with no
external heating and the lowest applied synchronized substrate bias,Vs¼ 20 V, have moduli E ¼ 346 GPa, a 40% increase with
respect to low-Ts DCMS TiN values. RaisingVsresults in an
additional increase inE to 443 GPa with Vs¼ 60 V and 497 GPa
withVs¼ 160 V. E does not increase significantly with further
increases inVsto 280 V.
The elastic moduli of Ti-HIPIMS/Ti-DCMS TiN films grown with metal–ion-synchronized Tiþ/Ti2þirradiation and Ts 120C exhibit a linear increase withVs (Fig. 7) from
279 GPa with Vs¼ 20 V to 388 GPa with Vs¼ 280 V, but
remain 25 to 35% lower than corresponding values for Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N layers deposited with
the sameVs.
Ta-HIPIMS/Ti-DCMS Ti1-xTaxN films with higher TaN
concentrations,x¼ 0.13 and 0.16, have higher hardness and elastic moduli values than Ti0.92Ta0.08N alloys for a given
Vs. H¼ 23 GPa for Ti0.87Ta0.13N grown with
metal–ion-synchronized Vs¼ 20 V pulses and increases to 27.8 and
30.8 GPa with Vs¼ 60 and 80 V. Correspondingly, E for
Ti0.87Ta0.13N layers increases from 435 to 540 and 611 GPa
withVs¼ 20, 60, and 80 V. H ¼ 26 GPa and E ¼ 517 GPa for
Ti0.84Ta0.16N layers grown with Vs¼ 20 V, increasing to
30.5 and 603 GPa with Vs¼ 60 V. However, the increase in
H and E values with increasing x is accompanied by higher compressive stresses, as shown in Sec.III E.
E. Residual stress
Residual stress values r, obtained from sin2w analyses, for Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N films are plotted as
a function of synchronized pulsed bias voltageVsin Fig.8.
Results are also shown for the four sets of reference layers described in Secs.IIandIII D.
The data presented in Fig.8are corrected for the tensile differential thermal contraction stress rth, which arises
dur-ing cooldur-ing of the samples from Ts (see Fig. 11 in the
Appendix) to room temperature, DT. rthis given by49
rth¼
E að f asÞDT
1 ; (3)
for which afand asare the film and substrate thermal
expan-sion coefficients.
The average thermal expansion coefficient of Si(001) over the temperature range from 120 to 23C is as¼ 2.9 106
K1.50 Since af for Ti1xTaxN films is unknown, we use a
linear extrapolation between aTiN¼ 9.35 106 K1 and
aTaN ¼ 8.00 106 K1 from Ref. 51 to obtain aTi0.92Ta0.08N
¼ 9.24 106 K1 for Ti0.92Ta0.08N. rth ranges from
0.25 GPa for Ti0.92Ta0.08N grown with Vs¼ 20 V to
0.35 GPa for Ti0.92Ta0.08N deposited with Vs¼ 280 V; for
reference DCMS and Ti-HIPIMS/Ti-DCMS layers grown with no external heating and the same Vs range, rth varies
from 0.23 to 0.32 GPa, while for the reference TiN film grown atTs¼ 500C withVs¼ 10 V, rth¼ 1.6 GPa.
FIG. 8. (Color online) Residual stress r for Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N/Si(001) films as a function of the applied substrate bias voltage
Vssynchronized to the metal–ion-rich portions of HIPIMS pulses. Results
are also shown for four sets of reference layers grown on Si(001) substrates: (1) DCMS TiN deposited with Vs¼ 10 V, (2) DCMS TiN grown at
Ts¼ 500C withVs¼ 60 V, (3) Ti-HIPIMS/Ti-DCMS TiN films deposited
with synchronized substrate bias 20 Vs 280 V, and (4) DCMS
Ti0.92Ta0.08N alloys grown atTs< 120C withVs¼ 10 V. Except for TiN
Underdense DCMS TiN reference layers grown with no external heating and Vs¼ 10 V are essentially stress-free,
r¼ 0.1 GPa, but DCMS TiN layers grown at 500C have
a compressive stress of 1.3 GPa. DCMS Ti0.92Ta0.08N
layers grown with no external heating andVs¼ 10 V also
ex-hibit low stress, r¼ 0.5 GPa. For Ti0.92Ta0.08N alloys
de-posited with no external heating and synchronous bias Vs 160 V, residual stresses are small, ranging from
r¼ 0.5 GPa with Vs¼ 20 V to 0.4 GPa with Vs¼ 80 V
to 0.7 GPa with Vs¼ 160 V. With Vs 200 V, r for
Ti0.92Ta0.08N increases to1.4 6 0.15 GPa.
Ti-HIPIMS/Ti-DCMS TiN films grown with Tiþ/Ti2þion irradiation, Vs 120 V, and no external heating are
essen-tially stress free. At higher synchronous biases, residual stresses become compressive and increase from 1.0 GPa withVs¼ 160 V to 1.2, 1.3, and 1.7 GPa with Vs¼ 200,
240, and 280 V.
Residual stresses in Ti1xTaxN alloys, deposited with no
external heating, increase with increasing TaN concentra-tion. r¼ 0.7 GPa for Ti0.87Ta0.13N films grown with
Vs¼ 20 V and increases rapidly to 1.6 and 2.1 GPa with
Vs¼ 60 and 80 V. Even higher stress values are obtained in
Ti0.84Ta0.16N layers, r¼ 1.4 for alloys deposited with
Vs¼ 20 V and 2.8 GPa with Vs¼ 60 V.
IV. DISCUSSION
DCMS TiN films deposited atTs< 120C (Ts/Tm 0.12)
withVs¼ 10 V exhibit pronounced inter- and intracolumnar
porosity (see XTEM and XSEM images in Figs.3and4) due to low adatom mobilities, which give rise to kinetic surface roughening52–54 exacerbated by atomic shadowing.55 The severe underdensity (65% dense, based upon XRR results) results in very low hardness,H¼ 7.8 GPa, and elastic modu-lus,E¼ 248 GPa, values.
The addition of pulsed heavy-metal Taþ/Ta2þion irradia-tion during growth of dilute Ti0.92Ta0.08N alloys in a hybrid
Ta-HIPIMS/Ti-DCMS mode, using a substrate bias applied synchronously with the metal–ion rich portions of the HIPIMS pulses, eliminates both inter- and intracolumn po-rosity due to effective near-surface atomic mixing allowing us to obtain essentially fully dense polycrystalline films. As a result, film hardness and elastic modulus increase dramati-cally. For Ti0.92Ta0.08N films grown with Vs¼ 160 V,
H¼ 25.9 GPa, and E ¼ 497 GPa, while the residual film stress remains low, r¼ 0.7 GPa, due to the short HIPIMS duty cycle (2%). Film surfaces are smoother [see Figs.3(b),
3(c), and 4] and pointed conical column tops are flattened. Further increases inVsdo not significantly enhanceH and E,
but slowly increase the residual compressive stress.
The effect of dilute alloying is minor compared to micro-structure modification via heavy-metal ion irradiation. This is clearly demonstrated by comparing the film hardness and elastic modulus results for DCMS TiN and Ti0.92Ta0.08N
layers grown with no applied heating (Ts< 120C),
Vs¼ 10 V, and no Ta ion bombardment (target material
ioni-zation during DCMS sputtering is negligible). H increases from 7.8 for DCMS TiN (Ts< 120C) to 9.8 GPa for DCMS
Ti0.92Ta0.08N (Ts< 120C), while E increases from 248 to
286 GPa as a result of dilute alloy formation. These increases are small compared to the effects of Taþ ion irradiation. Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N layers grown under
synchronized substrate biasVs¼ 160 V exhibit H ¼ 25.9 GPa
and E¼ 497 GPa, while the residual stress remains low (0.7 GPa).
Ti1xTaxN hardness can also be increased by raising the
TaN alloy concentration through increases in the peak HIPIMS current densityJTfor the same pulse width
(corre-sponding to higher pulse energies Ep). However, this comes
at the expense of increasing compressive stress.
Pulsed Tiþ/Ti2þmetal-ion irradiation during Ti-HIPIMS/ Ti-DCMS is not as effective as higher-mass Taþ/Ta2þ bombardment in densifying the growing film. Hence, Ti-HIPIMS/Ti-DCMS TiN layers have much lower hardnesses (e.g., H¼ 13.8 GPa with Vs¼ 160 V) than
Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N layers (H¼ 25.9 GPa)
grown with the sameVs. A crude first estimate of the
maxi-mum energy deposited to primary recoils by the incoming ion can be obtained by assuming 180 hard-sphere collisions of 160 eV Taþions (mTa¼ 180.95 amu) with Ti0.92Ta0.08N
(av-erage film mass mf¼ 36.26 amu) and Tiþ(mTi¼ 47.87 amu)
with TiN (mf¼ 30.94 amu). Under these conditions, the
maxi-mum fraction of incident Taþion energy deposited into the film isfTa¼ 4mTamf/(mTaþ mf)2¼ 0.56 and the corresponding
value for Tiþis fTi¼ 0.95. Thus, lower-mass Tiþ
bombard-ment creates higher-energy primary recoils (maximum recoil energy Er,max¼ 152 eV). A larger number of recoils with
lower energy (Er,max¼ 90 eV) are produced with Taþ
bom-bardment. This difference becomes even more pronounced for doubly charged metal ions, which constitute a significant fraction of the metal–ion flux incident at the growth surface (see Fig.1),Er,max¼ 304 eV for Ti2þversus 180 eV for Ta2þ.
A better estimate of energy deposition effects during film growth is obtained based upon Monte Carlo TRIM36
simula-tions. We characterize displacement cascades for heavier Ta ions versus lighter Ti ions, including ion ranges, depth profiles, and energy distributions of primary recoils, for 160 eV Ti bombardment of TiN and Ta bombardment of
TABLEI. TRIM simulation results for 160 eV Ti irradiation of TiN and Ta irradiation of Ti0.92Ta0.08N.Er;iare the average primary recoil energies and
nr;ivalues are the effective recoil depths fori¼ N, Ti, and Ta.
160 eV Ti 160 eV Ta N primary recoils/ion 1.4 1.1 Ti primary recoils/ion 1.8 2.0 Ta primary recoils/ion - 0.2
hEir;N(eV) 36 24
hEir;Ti(eV) 45 40
hEir;Ta (eV) - 52
nr;N(A˚ ) 12 13
nr;Ti(A˚ ) 13 16
nr;Ta(A˚ ) - 19
nr;Tiþ(A˚ ) 14
-nr;Taþ(A˚ ) - 25
Ti0.92Ta0.08N (see Table I). While TRIM is not accurate
enough at low ion energies for quantitative evaluation, it does provide useful qualitative insights. Taþcreates20% more primary cation recoils per incident ion, 2.0 Ti and 0.2 Ta versus 1.8 Ti recoils per ion for Tiþ, and20% fewer N recoils per ion, 1.1 versus 1.4.
Due to the large mass mismatch between Taþand N, only 22% of the incident ion energy is deposited into N primary recoils, while for Tiþ bombardment the fraction is 39%, almost twice as large. Moreover, there are important differ-ences in the energy distribution of the primary recoils for both cases, see Figs.9and10. For Taþirradiation, the N pri-mary recoil energy ETar;Nþis42 eV, while for Ti
þ, there is a
high energy ETir;Nþ tail extending to 110 eV. Similarly, a
high-energy component is present for Ti primary recoils under Tiþbombardment,ETir;Tiþ extends to 160 eV because of
the perfect mass match between ion and target atom. For Taþions, the maximum energy of Ti recoilsETar;Tiþ¼ 106 eV,
and there are a larger number of Ti recoils with energy 20–60 eV. Thus,TRIMresults indicate an increased number of
Ti primary recoils with lower recoil energies for Taþ/Ta2þ versus Tiþ/Ti2þions.
In addition, Taþ collision cascades extend over larger depths than these of Tiþ. The cascade for a given species is characterized by an effective depth nr, the sum of the
pro-jected rangerpand straggle Drp.
56
Results for Taþand Tiþ primary ions as well as Ti, Ta, and N recoils are given in Table I. The cascade generated by 160 eV Tiþ ions has a depth of13 A˚ for all three species: Ti primary ions and Ti and N recoils. With Taþbombardment, N recoils also extend to 13 A˚ , while the cation species, Ti and Ta, have longer ranges of 16 and 19 A˚ , respectively. The primary Taþions, because of their high mass, scatter at much smaller angles than Tiþand thus penetrate deeper, to 25 A˚ . Thus, both the primary Taþ metal ions and the recoiled Ta lattice atoms penetrate deeper into the near-surface region to dynamically fill residual vacancies, which result from very-low-tempera-ture growth (Ts/Tm < 0.12). Ta-HIPIMS with synchronous
metal–ion bias is much more effective in film densification than Ti-HIPIMS, as shown by XTEM (Fig. 3) and XRR results, giving rise to higher hardness and elastic modulus values with—depending on the choice of HIPIMS duty cycle, ion flux, ion charge, and substrate bias—low residual film stress.
In order to achieve densification, two conditions have to be satisfied simultaneously: (1) the ion dose and (2) the ion penetration range nion have to be sufficiently high that the
extent of the near-surface intermixing zone, defined by the average effective depth of collision cascades, nmix 10–30 A˚
(depending on the mass and energy of the collision partners), is larger than the DCMS layer thickness deposited between successive HIPIMS pulses. Moreover, the number of recoils created per metal atom deposited between the pulses has to be sufficiently high to ensure adequate adatom displacements in the newly deposited DCMS layer to achieve full densification. In the present Ti0.92Ta0.08N experiments, the TiN thickness
deposited between pulseshTiNis 0.022 ML, while the TiN and
the TaN thicknesses deposited during a HIPIMS pulse are 0.00045 and 0.002 ML, respectively. Thus, the real-time TaN deposition rate is 9 higher than that of TiN; however, the time-averaged rate is 11 lower. Dynamic near-surface mix-ing due to Taþ/Ta2þion irradiation during HIPIMS pulses is very effective for two reasons: (1) nmix hTiN, and (2) the
high real-time TaN deposition rate leads to strongly overlap-ping collision cascades.
With x¼ 0.08, metal–ion-synchronized Vs¼ 160 V, a
duty cycle of 2%,Ts 120C, andhTiN¼ 0.022 ML, residual
stresses in Ti0.92Ta0.08N films are low. This is predominantly
due to the fact that irradiation by gas ions is suppressed due to the synchronous biasing.21 Moreover, the metal bombarding ions are film constituents and primarily incorporated into lat-tice sites, consistent with the increased relaxed latlat-tice parame-ter [see Fig. 2(b) and Sec. III C], resulting in relatively low residual defect concentrations. The above benefits of pulsed heavy-ion irradiation during very low-temperature film growth cannot be achieved by bombardment with heavy gas ions such as Xeþ (mXe¼ 131.29 amu) since a sufficient flux to cause
densification will also result in a high compressive stress.57 Further increase in the incident Taþ/Ta2þflux, hence the TaN film concentrationx, while maintaining the TiN growth
FIG. 9. (Color online) Energy distributions, obtained from TRIM simula-tions, of primary N recoils due to 160 eV Ta irradiation of Ti0.92Ta0.08N and
160 eV Ti irradiation of TiN.
FIG. 10. (Color online) Energy distributions, obtained from TRIM
simula-tions, of primary Ti recoils due to 160 eV Ta irradiation of Ti0.92Ta0.08N and
rate constant, is not a preferred strategy for increasing film density,H, and E, as it leads to higher residual stress values (see Sec.III E).
The densification mechanism demonstrated here is not limited to the addition of Ta ions. Other group-VI transition metals with sufficiently high mass and low ionization poten-tial, comparable to that of Ta (m¼ 180.95 amu and IP1
¼ 7.91 eV), to ensure significant ionization during HIPIMS pulses, such as Hf (m¼ 178.49 amu and IP1¼ 6.85 eV) or W
(m¼ 183.84 amu, IP1¼ 8.01 eV), would provide the same
benefits. In addition, the pulsed heavy-metal–ion irradiation process demonstrated here can also be applied to multica-thode systems in which several targets operate in DCMS mode to provide high deposition rates, while heavy-metal– ion bombardment is supplied from one or more HIPIMS sources.
V. CONCLUSIONS
We demonstrate a new process to obtain fully dense re-fractory ceramic thin films with high hardness and elastic modulus, and low residual stress, atTsnear room
tempera-ture (Ts/Tm< 0.12). The procedure is based upon reactive
hybrid HIPIMS/DCMS co-sputtering in which the substrate bias is applied synchronously with the metal–ion-rich por-tions of the HIPIMS pulses. The primary target, Ti in the present example, operates in DCMS mode providing a con-tinuous flux of sputter-ejected metal atoms to sustain a high deposition rate, while a high-mass/low-IP1target metal, Ta,
is driven by HIPIMS to serve as a pulsed source of energetic metal–ions to irradiate the growth surface and densify dilute Ti1xTaxN alloy layers. No external heating is used. As a
result of heavy-metal–ion bombardment, the inter- and intra-columnar porosity, typical of low-deposition-temperature re-fractory ceramic thin film growth is eliminated due to effective near-surface atomic mixing, as evidenced by XTEM, XSEM, and XRR results. The upper limit for the DCMS layer thickness grown between successive HIPIMS pulses is set by the depth of the near-surface mixing zone nmix, which is controlled by the choice of the film elemental
masses and the metal–ion energy.
For the model Ti0.92Ta0.08N alloy films, grown by
Ta-HIPIMS/Ti-DCMS at Ts 120C with the bias voltage
Vs¼ 160 V synchronized to the Taþ/Ta2þ portions of the
HIPIMS pulses, discussed in this article, film hardness and elastic modulus are 330 and 200% higher than corresponding values for reference DCMS TiN (Vs¼ 10 V) layers, while
the residual stress remains low. The effect of dilute alloying on enhanced mechanical properties of Ta-HIPIMS/ Ti-DCMS Ti0.92Ta0.08N, relative to DCMS TiN, films is
small as compared to the effect of microstructure densifica-tion caused by heavy-metal Taþ/Ta2þion irradiation. Pulsed Tiþ/Ti2þmetal–ion irradiation during Ti-HIPIMS/Ti-DCMS is not nearly as effective as higher-mass Taþ/Ta2þ bombard-ment in densifying the growing film. This is interpreted with the help of TRIM simulations, which show an increased number of Ti lattice-atom primary recoils with lower recoil energies, and deeper collision cascades, for Taþ/Ta2þversus
Tiþ/Ti2þ ions. Additional lattice relaxation is provided by the high real-time ion flux, during the pulses, resulting in strongly overlapping collision cascades.
Residual defect concentrations are low under Tanþ bom-bardment, since Ta is primarily incorporated into the lattice of the growing alloy film, giving rise to an increased relaxed lattice parameter, resulting in low residual stress values. These results are not possible to achieve by bombardment with heavy noble-gas ions such as Xeþ(mXe¼ 131.29 amu),
of similar energy, which become incorporated as interstitials. High hardness and low residual stress obtained for Ta-HIPIMS/Ti-DCMS Ti0.92Ta0.08N alloys grown atTs 120C
indicate that the heavy-metal ion momentum transfer to the growing film provides sufficient adatom mobility and near-surface intermixing at Ts near RT to substitute for the
thermally driven adatom diffusion even atTs¼ 500C.
ACKNOWLEDGMENTS
Financial support from the European Research Council (ERC) through an Advanced Grant, the VINN Excellence Center Functional Nanoscale Materials (FunMat), the Knut and Alice Wallenberg Foundation, and the Strategic Faculty Grant in Materials Science (AFM) is gratefully acknowl-edged. The authors thank Fredrik Eriksson and Jens Jensen for help with XRR and RBS measurements, respectively. APPENDIX: TEMPERATURE RISE DURING FILM GROWTH
Figure11 shows substrate temperatures Ts as a function
of timet during film growth, with no externally applied sub-strate heating, of Ta-HIPIMS/Ti-DCMS Ti1xTaxN alloys
deposited using a metal–ion-synchronized pulsed bias Vs¼ 160 V, and DCMS TiN films with Vs¼ Vf¼ 10 V.
Ti1xTaxN results are shown for three different TaN
concen-trations: x¼ 0.08 (Ep¼ 5 J, JT¼ 0.21 A/cm2), x¼ 0.13 (Ep¼ 10 J, JT¼ 0.53 A/cm 2 ), and x¼ 0.16 (Ep¼ 15 J, JT¼ 0.93 A/cm 2
. The starting temperature in all cases, fol-lowing thermal degassing of the Si(001) substrates prior to film deposition, isTs¼ 65 6 1C.
FIG. 11. (Color online) Substrate temperaturesTsas a function of deposition
time t during growth of Ta-HIPIMS/Ti-DCMS Ti1xTaxN (x¼ 0.08, 0.13,
and 0.16) with Vs¼ 160 V synchronized to the metal–ion-rich portions of
HIPIMS pulses, and DCMS TiN films grown withVs¼ 10 V. In all cases,
there is no external heating of the Si(001) substrates.
All Ts(t) curves are composed of two distinct regions.
Ts(t) increases rapidly during the first10 min with a
heat-ing rate dTs/dt of 2.6C/min for TiN and 3.1C/min for
Ti0.92Ta0.08N deposition. The initial heating rate dTs/dt
increases to 3.7 and 4.3C/min for Ti1xTaxN alloys with
x¼ 0.13 and 0.16. In all cases, dTs/dt decreases at longer
deposition times and becomes nearly constant for the final 30 min at 0.32C/min for TiN and 0.33C/min for Ti0.92Ta0.08N, increasing to 0.4C/min for alloys with
x¼ 0.13 and 0.43C/min with x¼ 0.16. For the reference DCMS TiN film,Tsreaches 118C during 60 min deposition
runs, while for Ti0.92Ta0.08N,Tsreaches 120C.
1
D. M. Mattox and G. J. Kominiak,J. Vac. Sci. Technol.9, 528 (1972).
2
G. Hakansson, J.-E. Sundgren, D. McIntyre, J. E. Greene, and W.-D. Munz,Thin Solid Films153, 55 (1987).
3I. Petrov, P. B. Barna, L. Hultman, and J. E. Greene,J. Vac. Sci. Technol.
A21, S117 (2003).
4
L. Hultman, J.-E. Sundgren, J. E. Greene, D. B. Bergstrom, and I. Petrov,
J. Appl. Phys.78, 5395 (1995).
5F. Adibi, I. Petrov, J. E. Greene, L. Hultman, and J.-E. Sundgren,J. Appl.
Phys.73, 8580 (1993).
6
T. C. Huang, G. Lim, F. Parmigiani, and E. Kay,J. Vac. Sci. Technol. A
3, 2161 (1985).
7
J. E. Greene,Deposition Technologies for Films and Coatings, edited by R. F. Bunshah (Noyes, Park Ridge, NJ, 1994), p. 681.
8
A. Anders, Cathodic Arcs, Springer Series on Atomic, Optical, and Plasma Physics Vol. 50 (Springer Science+Business Media, LLC, New York, 2009).
9
V. Kouznetsov, K. Macak, J. M. Schneider, U. Helmersson, and I. Petrov,
Surf. Coat. Technol.122, 290 (1999).
10T. Lee, H. Seo, H. Hwang, B. Howe, S. Kodambaka, J. E. Greene, and
I. Petrov,Thin Solid Films518, 5169 (2010).
11
J. E. Greene, “Thin Film Nucleation, Growth: Microstructural Evolution: An Atomic Scale View,” in Handbook of Deposition Technologies for Thin Films and Coatings, 3rd ed., edited by P. Martin (William Andrew, Burlington, MA, 2010).
12
H. Oettel, R. Wiedemann, and S. Preisler,Surf. Coat. Technol.74–75, 273 (1995).
13C. V. Falub, A. Karimi, M. Ante, and W. Kalss,Surf. Coat. Technol.201,
5891 (2007).
14
A. H€orling, L. Hultman, M. Oden, J. Sj€olen, and L. Karlsson,J. Vac. Sci. Technol. A20, 1815 (2002).
15J. A. Thornton,J. Vac. Sci. Technol. A4, 3059 (1986).
16J. M. E. Harper, J. J. Cuomo, and R. J. Gambino, Ion Bombardment
Modification of Surfaces: Fundamentals and Applications, edited by O. Auciello and R. Kelly (Elsevier, Amsterdam, 1984).
17H. F. Winters and E. Kay,J. Appl. Phys.38, 3928 (1967). 18J. A. Thornton, J. Tabock, and D. W. Hoffman,Thin Solid Films
64, 111 (1979).
19
G. Greczynski, J. Lu, M. Johansson, J. Jensen, I. Petrov, J. E. Greene, and L. Hultman,Surf. Coat. Technol.206, 4202 (2012).
20
G. Greczynski, J. Lu, M. Johansson, J. Jensen, I. Petrov, J. E. Greene, and L. Hultman,Vacuum86, 1036 (2012).
21
G. Greczynskiet al.,J. Vac. Sci. Technol. A30, 061504 (2012).
22V. Petrman and J. Houska,J Mater Sci.48, 7642 (2013).
23A. M. James and M. P. Lord,Macmillan’s Chemical and Physical Data
(Macmillan, London, UK, 1992).
24
J. E. Huheey, E. A. Keiter, and R. L. Keiter, Inorganic Chemistry: Principles of Structure and Reactivity, 4th ed. (HarperCollins, New York, USA, 1993).
25CRC Handbook of Chemistry and Physics, 84th ed., edited by David R.
Lide, Section 10, Atomic, Molecular, and Optical Physics; Ionization Potentials of Atoms and Atomic Ions (CRC, Boca Raton, FL, 2003).
26
See: http://www.cemecon.de/coating_technology/coating_units/hipims_ sputter_coating_system/index_eng.html, accessed in June 2014.
27Thermocouple calibrations are confirmed using boiling water; the
maxi-mum allowable deviation is 60.2C.
28
Current densities reported here are determined based upon the full flat-target area. Current densities based upon the shape of the erosion pattern following the experiments are estimated as 0.31 A/cm2withE
p¼ 5 J, 0.82
A/cm2withEp¼ 10 J, and 1.49 A/cm2withEp¼ 15 J. 29
G. Greczynski and L. Hultman,Vacuum84, 1159 (2010).
30M. Mayer, AIP Conf. Proc. 475, 541 (1999).
31M. Birkholz, in Thin Film Analysis by X-ray Scattering (Wiley-VCH,
Weinheim 2006), Chap. 6.
32
J. O. Kim, J. D. Achenbach, P. B. Mirkarimi, M. Shinn, and S. A. Barnett,
J. Appl. Phys.72, 1805 (1992).
33P. Djemia, M. Benhamida, Kh. Bouamama, L. Belliard, D. Faurie, and G.
Abadias,Surf. Coat. Technol.215, 199 (2013).
34
W. C. Oliver and G. M. Pharr,J. Mater. Res.7, 1564 (1992).
35J. F. Ziegler, J. P. Biersack, and U. Littmark, “The stopping and range
of ions in solids,” inStopping and Ranges of Ions in Matter (Pergamon, New York, 1984), Vol. 1.
36
See:www.srim.org; accessed on 28 June 2013.
37The JCPDS database (1998), data set number: 38-1420; W. Wong-Ng, H.
McMurdie, B. Paretzkin, C. Hubbard, and A. Dragoo,Powder Diffr. 2, 191 (1987).
38
C.-S. Shin, D.Gall, Y.-W. Kim, P. Desjardins, I. Petrov, J. E. Greene, M. Oden, and L. Hultman,J. Appl. Phys.90, 2879 (2001).
39G. Ehrlich and F. G. Hudda,J. Chem. Phys.
44, 1039 (1966).
40
G. Ehrlich,Surf. Sci.331–333, 865 (1995).
41
G. K. Wehner and D. Rosenberg,J. Appl. Phys.31, 177 (1960).
42P. Sigmund,Phys. Rev.184, 383 (1969). 43P. Sigmund,Phys. Rev.
187, 768 (1969).
44
T. Ono, T. Keamotsu, and T. Muramoto, Reactive Sputter Deposition, edited by D. Depla and S. Mahieu (Springer, Berlin, Germany, 2008).
45A. M. Myers, J. R. Doyle, and D. N. Ruzic, J. Appl. Phys.72, 3064
(1992).
46
L. Hultman, U. Helmersson, S. A. Barnett, J.-E. Sundgren, and J. E. Greene,J. Appl. Phys.61, 552 (1987).
47H. Ljungcrantz, M. Oden, L. Hultman, J. E. Greene, and J.-E. Sundgren,
J. Appl. Phys.80, 6725 (1996).
48
C.-S. Shin, D. Gall, N. Hellgren, J. Patscheider, I. Petrov, and J. E. Greene,J. Appl. Phys.93, 6025 (2003).
49M. Ohring, Materials Science of Thin Films, 2nd ed. (Academic, New
York, 2001), pp. 732–733.
50
H. Watanabe, N. Yamada, and M. Okaji, Int. J. Thermophys.25, 221 (2004).
51Hugh O. Pierson,Handbook of Refractory Carbides and Nitrides: Properties,
Characteristics, Processing, and Applications (William Andrew, Burlington, MA, 1996). p. 193.
52
N.-E. Lee, D. G. Cahill, and J. E. Greene,J. Appl. Phys.80, 2199 (1996).
53J. Krug, P. Politi, and T. Michely,Phys. Rev. B61, 14037 (2000). 54
B. W. Karr, I. Petrov, D. G. Cahill, and J. E. Greene,Appl. Phys. Lett.70, 1703 (1997).
55
C. Roland and H. Guo,Phys. Rev. Lett.66, 2104 (1991).
56E. F. Krimmel and H. Pfleiderer,Radiat. Effects19, 83 (1973).
57L. Hultman, C. Engstrom, J. Birch, M. P. Johansson, M. Oden, L.